Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
2001-09-24
2004-01-06
Britton, Howard (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C348S413100, C348S699000, C375S240120
Reexamination Certificate
active
06674798
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to devices for detecting motion vectors which are employed for motion compensation of a moving picture image in predictive coding.
2. Description of the Background Art
A data compression technique of reducing an amount of data is indispensable for transmitting or storing picture signals having a large amount of data. Picture data have considerable redundancy resulting from correlation between adjacent pixels and human perceptional characteristics. A data compression technique of suppressing the data redundancy for reducing the volume of data for transmission of data is called high efficiency coding. One of such high efficiency coding systems is a frame-to-frame predictive coding system, which is adapted to carry out the following processing:
A predictive error, which is a difference between each pixel data of a current screen (frame or field) to be currently coded and each pixel data of the same position of a precedent screen to be referred to is calculated, so that the predictive error as calculated is thereafter employed for coding. According to this method, it is possible to code picture images having small movements in high efficiency, due to high correlation between the screens. As to picture images having large movements, however, errors are disadvantageously increased due to small correlation between screens, leading to increase in volume of data to be transmitted.
In order to solve the aforementioned problem, a frame-to-frame (field-to-field) predictive coding system with motion compensation is adapted to carry out the following processing: First, motion vectors are previously calculated through pixel data of a current screen (frame or field) and a precedent screen before calculating predictive errors. A predictive picture image of the precedent screen is moved in accordance with the motion vector as calculated. Picture data in a position which is displaced from that of the precedent screen by the motion vector are regarded as reference pixels, which in turn are employed as predicted values. Then, predictive errors between respective pixels of the precedent screen as moved and the current screen are calculated so that the predictive errors and the motion vectors are transmitted.
FIG. 151
is a block diagram schematically showing an overall structure of a conventional encoder for coding picture data in accordance with the predictive coding system with motion compensation. Referring to
FIG. 151
, the encoder includes a preprocessing circuit
910
for carrying out prescribed preprocessing on picture signals as received, a source coding circuit
912
for eliminating redundancy from the signals preprocessed by the preprocessing circuit
910
and quantizing input signals, and a video multiplex coding circuit
914
for coding signals received from the source coding circuit
912
in accordance with a prescribed format and multiplexing the coded signals to a code train of a predetermined data structure.
The preprocessing circuit
910
converts input picture signals to those of a common intermediate format (CIF) through time and space filters, and performs filter processing for noise removal.
The source coding circuit
912
performs orthogonal transformation processing such as discrete cosine transformation (DCT) on received signals as well as motion compensation for input signals, while quantizing picture data subject to the orthogonal transformation.
The video multiplex coding circuit
914
performs two-dimensional variable-length coding on received picture signals with variable-length coding of various attributes, such as motion vectors, of blocks which are units of data processing, and thereafter multiplexes the signals to a code train of a predetermined data structure.
The encoder further includes a transmission buffer
916
for buffering picture data from the video multiplex coding circuit
914
, and a transmission coding circuit
918
for adapting the picture data from the transmission buffer
916
to a transmission channel.
The transmission buffer
916
smooths information generating speeds to a constant speed. The transmission coding circuit
918
executes addition of error checking bits and sound signal data.
FIG. 152
illustrates an exemplary structure of the source coding circuit
912
shown in FIG.
151
. Referring to
FIG. 152
, the source coding circuit
912
includes a motion compensation predictor
920
for detecting motion vectors with respect to input picture signals received from the preprocessing circuit
910
and forming reference picture images motion-compensated in accordance with the motion vectors, a loop filter
922
for performing filter processing on reference picture image pixel data received from the motion compensation predictor
920
, a subtracter
924
for obtaining differences between outputs of the loop filter
922
and input picture signals, an orthogonal transformer
926
for orthogonally transforming outputs of the subtracter
924
, and a quantizer
928
for quantizing data orthogonally transformed by the orthogonal transformer
926
.
The motion compensation predictor
920
, the structure of which is described later in detail, includes a frame memory for storing pixel data of a precedent frame, for detecting motion vectors and forming motion-compensated reference picture image pixel data in accordance with input picture signal data and pixel data in the frame memory. The loop filter
922
is provided to improve the picture quality.
The orthogonal transformer
926
carries out orthogonal transformation such as DCT transformation on data received from the subtracter
924
in units of blocks of a prescribed size (8 by 8 pixels in general). The quantizer
928
quantizes the orthogonally transformed pixel data.
The motion compensation predictor
920
and the subtracter
924
execute frame-to-frame prediction with motion compensation, for eliminating time redundancy in a motion image. Further, spatial redundancy in motion image signals is eliminated by orthogonal transformation through the orthogonal transformer
926
.
The source coding circuit
912
further includes an inverse quantizer
930
for transforming the data quantized in the quantizer
928
to the original signal states, an inverse orthogonal transformer
932
for performing inverse orthogonal transformation on outputs of the inverse quantizer
930
, and an adder
934
for adding up outputs of the loop filter
922
and the inverse orthogonal transformer
932
. The inverse quantizer
930
and the inverse orthogonal transformer
932
form a picture image which is employed in frame-to-frame prediction for a subsequent frame. The picture data as generated are written in the frame memory which is included in the motion compensation predictor
920
. The adder
934
adds picture signals (frame-to-frame difference data) to the outputs of the loop filter
922
, whereby the picture data of the current frame are reproduced. In general, such inverse quantization, inverse orthogonal transformation and addition are called local decoding processes. Calculation of the motion vectors is now described more specifically. In general, a block matching method is employed for calculating the motion vectors.
As shown in
FIG. 153A
, consider that a picture image A in a (m−1)-th frame is moved to A′ in an m-th frame. In the block matching method, the screen (one frame in this case) is divided into blocks each including P by Q pixels (P=Q in general). A precedent frame is searched for a block which is most approximate to that of interest in the current frame. Displacement from the interested block to the most approximate block in the precedent frame is called a motion vector. Description is now made in more detail.
As shown in
FIG. 153B
, it is assumed that the m-th frame is to be coded. The frame is divided into blocks each having N by N pixels (P=Q=N). It is assumed that pixel data in the upper leftmost pixel position (Nk, N1) in the block of the N by N pixels in the m-th frame has a value X
Hanami Atsuo
Ishihara Kazuya
Kumaki Satoshi
Matsumura Tetsuya
Nakagawa Shin-ichi
Britton Howard
McDermott & Will & Emery
Renesas Technology Corp.
LandOfFree
Motion vector detecting device capable of accommodating a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Motion vector detecting device capable of accommodating a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Motion vector detecting device capable of accommodating a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3223382